Making multiple HTTP requests asynchronously

问题:

require 'net/http'

urls = [
  {'link' => 'http://www.google.com/'},
  {'link' => 'http://www.yandex.ru/'},
  {'link' => 'http://www.baidu.com/'}
]

urls.each do |u|
  u['content'] = Net::HTTP.get( URI.parse(u['link']) )
end

print urls

This code works in synchronous style. First request, second, third. I would like to send all requests asynchronously and print urls after all of them is done.

What the best way to do it? Is Fiber suited for that?

回答1:

Here's an example using threads.

require 'net/http'

urls = [
  {'link' => 'http://www.google.com/'},
  {'link' => 'http://www.yandex.ru/'},
  {'link' => 'http://www.baidu.com/'}
]

urls.each do |u|
  Thread.new do
    u['content'] = Net::HTTP.get( URI.parse(u['link']) )
    puts "Successfully requested #{u['link']}"

    if urls.all? {|u| u.has_key?("content") }
      puts "Fetched all urls!"
      exit
    end
  end
end

sleep

回答2:

I just saw this, a year and a bit later, but hopefully not too late for some googler...

Typhoeus by far the best solution for this. It wraps libcurl in a really elegant fashion. You can set the max_concurrency up to about 200 without it choking.

With respect to timeouts, if you pass Typhoeus a :timeout flag, it will just register a timeout as the response... and then you can even put the request back in another hydra to try again if you like.

Here's your program rewritten with Typhoeus. Hopefully this helps anybody who comes across this page later!

require 'typhoeus'

urls = [
  'http://www.google.com/',
  'http://www.yandex.ru/',
  'http://www.baidu.com/'
]

hydra = Typhoeus::Hydra.new

successes = 0

urls.each do |url|
    request = Typhoeus::Request.new(url, timeout: 15000)
    request.on_complete do |response|
        if response.success?
            puts "Successfully requested " + url
            successes += 1
        else
            puts "Failed to get " + url
        end
    end
    hydra.queue(request)
end

hydra.run 

puts "Fetched all urls!" if successes == urls.length

回答3:

I have written an in-depth blog post about this topic which includes an answer that is somewhat similar to the one August posted - but with a few key differences: 1) Keeps track of all thread references in "thread" array. 2) Uses "join" method to tie up threads at the end of program.

require 'net/http'

# create an array of sites we wish to visit concurrently.
urls = ['link1','link2','link3']  
# Create an array to keep track of threads.
threads = []

urls.each do |u|  
  # spawn a new thread for each url
  threads << Thread.new do
  Net::HTTP.get(URI.parse(u))
    # DO SOMETHING WITH URL CONTENTS HERE
    # ...
    puts "Request Complete: #{u}\n"
  end
end

# wait for threads to finish before ending program.
threads.each { |t| t.join }

puts "All Done!"  

The full tutorial (and some performance information) is available here: https://zachalam.com/performing-multiple-http-requests-asynchronously-in-ruby/

回答4:

This can be done with the C library cURL. A ruby binding for that library exists, but it doesn't seem to support this functionality out of the box. However, it looks like there is a patch adding/fixing it (example code is available on the page). I know this doesn't sound great, but it might be worth a try if there aren't any better suggestions.

回答5:

It depends what you want to do after the function afterwards. You can do it with simple threads:

see: http://snipplr.com/view/3966/simple-example-of-threading-in-ruby/

回答6:

You could have a different thread execute each one of the Net::HTTP.get. And just wait for all the threads to finish.

BTW printing urls will print both the link and the content.

回答7:

The work_queue gem is the easiest way to perform tasks asynchronously and concurrently in your application.

wq = WorkQueue.new 2 # Limit the maximum number of simultaneous worker threads

urls.each do |url|
  wq.enqueue_b do
    response = Net::HTTP.get_response(url)
    # use the response
  end
end

wq.join # All requests are complete after this